as_finite_diff for complex PDE

[Mathias Louboutin]

Hi

I am reposting this as I didn't get any answer the last time. I am working on discret wave equations and I need a symbolic expression for 

$\frac{d^2 p(x,t)}{dt^2} - G_{\bar{x}\bar{x}}^T((1+2\epsilon) p(x,t)) - G_{\bar{z}\bar{z}}^T(\sqrt{(1+2\delta)} r(x,t)) =q$

where

$G_{\bar{x}\bar{x}} & = cos(\theta)^2 \frac{d^2}{dx^2} + sin(\theta)^2 \frac{d^2}{dz^2} - sin(2\theta) \frac{d^2}{dx dz} \\

G_{\bar{z}\bar{z}} & =  sin(\theta)^2 \frac{d^2}{dx^2} + cos(\theta)^2 \frac{d^2}{dz^2} +sin(2\theta) \frac{d^2}{dx dz} \\ $

however as_finite_diff doesn't allow any cross derivatives (d/dxdy) even so the documentation says it does (it only does if you define it and then only take d/dx or d/dy)

The second problem is to take derivatives of product. It shouldn't be a problem as this is just another function but it doesn't work neither.

thank you

mathias louboutin

[Björn Dahlgren]

On Tuesday, 12 April 2016 12:02:40 UTC+2, Mathias Louboutin wrote:

however as_finite_diff doesn't allow any cross derivatives (d/dxdy) even so the documentation says it does (it only does if you define it and then only take d/dx or d/dy)

I opened an issue for this: https://github.com/sympy/sympy/issues/11007

[Mathias Louboutin]

Sorry for the late answer. Thank you for opening the issue I'll check it out.